Re: Extents: Less or More?

From: Phil Herring <revdoc_at_uow.edu.au>
Date: 1998/02/03
Message-ID: <6b89dj$bos$1@wyrm.its.uow.edu.au>#1/1

In article <34D69C83.5CE5_at_p3.net> Jerry Gitomer, jgitomer_at_p3.net writes:
>I had the misfortune to get nailed by an excessive number of extents
>problem in a purchased application. The application was written in
>a manner that caused a few of the columns in the row to be entered
>initially and then added additional columns at separate times over
>the course of the day. By the end of the day the main table had
>grown from one extent to seventeen :-(

This problem isn't due to an excessive number of extents, it's due to chained rows (which are almost always a Very Bad Thing Indeed). This results in single rows being split up into pieces located all over the disk, so that getting one row takes many more I/Os than it should.

In cases like this, you need to attend to PCTFREE and PCTUSED, rather than the number of extents per se, although as you found, an export/import cycle will also fix the problem temporarily.

Received on Tue Feb 03 1998 - 00:00:00 CST